Skip to Main Content
ASME Press Select Proceedings

International Conference on Software Technology and Engineering (ICSTE 2012)

Editor
Jianhong Zhou
Jianhong Zhou
Search for other works by this author on:
ISBN:
9780791860151
No. of Pages:
680
Publisher:
ASME Press
Publication date:
2012

Financial time series forecasting has long been an application area for soft computing techniques. The essential problem is to find a soft computing model that is compatible with technical analysis models used in stock markets. Backpropagation neural networks have often been used to emulate a technical analyst’s prediction. In this paper, two primary technical indicators, William’s %R and Moving Average Convergence Divergence (MACD), have been used for feature extraction from price data. We have focused on optimizing the separabilty of the resulting feature space. Levenberg-Marquardt variation of backpropagation algorithm has been found to be an effective decision engine to predict stock market trends. Due to the complexity of input data feature space, this algorithm converges to sub-optimal minima. Hence, genetic algorithms have been identified as an effective optimization tool to converge closer to global minimum. Accuracy using this method is found to be 50% to 80% for a number of stocks listed on the Bombay Stock Exchange (BSE).

This content is only available via PDF.
Close Modal
This Feature Is Available To Subscribers Only

Sign In or Create an Account

Close Modal
Close Modal